Tamara Rordorf-Nikolic

We propose a novel model for the regulation of the p85/pl10alpha phosphatidylinositol 3'-kinase. In insect cells, the p110alpha catalytic subunit is active as a monomer but its activity is decreased by coexpression with the p85 regulatory subunit. Similarly, the lipid kinase activity of recombinant glutathione S-transferase (GST)-p110alpha is reduced by 65 to 85% upon in vitro reconstitution with p85. Incubation of p110alpha/p85 dimers with phosphotyrosyl peptides restored activity, but only to the level of monomeric p110alpha. These data show that the binding of phosphoproteins to the SH2 domains of p85 activates the p85/p110alpha dimers by inducing a transition from an inhibited to a disinhibited state. In contrast, monomeric p110 had little activity in HEK 293T cells, and its activity was increased 15- to 20-fold by coexpression with p85. However, this apparent requirement for p85 was eliminated by the addition of a bulky tag to the N terminus of p110alpha or by the growth of the HEK 293T cells at 30 degrees C. These nonspecific interventions mimicked the effects of p85 on p110alpha, suggesting that the regulatory subunit acts by stabilizing the overall conformation of the catalytic subunit rather than by inducing a specific activated conformation. This stabilization was directly demonstrated in metabolically labeled HEK 293T cells, in which p85 increased the half-life of p110. Furthermore, p85 protected p110 from thermal inactivation in vitro. Importantly, when we examined the effect of p85 on GST-p110alpha in mammalian cells at 30 degrees C, culture conditions that stabilize the catalytic subunit and that are similar to the conditions used for insect cells, we found that p85 inhibited p110alpha. Thus, we have experimentally distinguished two effects of p85 on p110alpha: conformational stabilization of the catalytic subunit and inhibition of its lipid kinase activity. Our data reconcile the apparent conflict between previous studies of insect versus mammalian cells and show that p110alpha is both stabilized and inhibited by dimerization with p85.

Phosphatidylinositol 3'-kinase (PI 3'-kinase) is activated in insulin-stimulated cells by the binding of the SH2 domains in its 85-kDa regulatory subunit to insulin receptor substrate-1 (IRS-1). We have previously shown that both tyrosyl-phosphorylated IRS-1 and mono-phosphopeptides containing a single YXXM motif activate PI 3'-kinase in vitro. However, activation by the monophosphopeptides was significantly less potent than activation by the multiply phosphorylated IRS-1. We now show that the increased potency of PI 3'-kinase activation by IRS-1 relative to phosphopeptide is not due to tertiary structural features IRS-1, as PI 3'-kinase is activated normally by denatured, reduced, and carboxymethylated IRS-1. Furthermore, activation of PI 3'-kinase by bis-phosphorylated peptides containing two YXXM motifs is 100-fold more potent than the corresponding mono-phosphopeptides and similar to activation by IRS-1. These data suggest that tyrosyl-phosphorylated IRS-1 or bis-phosphorylated peptides bind simultaneously to both SH2 domains of p85. However, these data cannot differentiate between an activation mechanism that requires two-site occupancy for maximal activity as opposed to one in which bivalent binding enhances the occupancy of a single activating site. To distinguish between these possibilities, we produced recombinant PI 3'-kinase containing either wild-type p85 or p85 mutated in its N-terminal, C-terminal, or both SH2 domains. We find that mutation of either SH2 domains significantly reduced phosphopeptide binding and decreased PI 3'-kinase activation by 50%, whereas mutation of both SH2 domains completely blocked binding and activation. These data provide the first direct evidence that full activation of PI 3'-kinase by tyrosylphosphorylated proteins requires occupancy of both SH2 domains in p85.